1. Field
The present invention is related to wireless communications systems.
2. Description of the Related Art
Conventional networks manage radio resources according to criteria established by the network. This may include a fixed allocation strategy or a dynamic allocation strategy. A core radio resource management (RRM) element of a commercial wireless network such as a Long Term Evolution (LTE) network may include a scheduler. Such a scheduler may generate a schedule for radio resources at a given wireless network infrastructure base station (e.g. in LTE, an evolved Node B or eNB), based on a multiplicity of parameters reflecting allocation criteria. Parameters input to such a scheduler may include Backlog[1 . . . Nue], QoS[1 . . . Nue], CQI[1 . . . Nue], Nue, UEMode[1 . . . Nue], Nprb, PRB[1 . . . Nprb][1 . . . Nsfr], ANT[1 . . . Na], and UEant[1 . . . Nue].
The Backlog[1 . . . Nue] may include the amount of data backlogged to each of hundreds of user equipment (UE) being serviced by such an eNB. The quality of service (QoS), if any, may include the QoS requirement parameters assigned to each such UE such as minimum bit rate and whether the traffic needs guaranteed bit rate (GBR) or not. The channel quality indication (CQI) may include the CQI of each such UE. Nue may include the number of UE currently known to the eNB. The UEMode may include the mode of each UE and may be expressed as a binary matrix indicating whether UE is in idle or connected mode. The Nprb may include the number of physical (radio) resource blocks (PRBs) available to the eNB. The PRB[1 . . . Nprb][1 . . . Nsfr] may represent the instantaneous usage of PRB in the scheduled LTE frames and subframes among Nsfr subframes being scheduled. The ANT[1 . . . Na] may represent the eNB antenna capabilities, indicating degree of multiple input multiple output (MIMO) available in each of Na cells and elements. The UEant[1 . . . Nue] may represent the antenna capabilities indicating MIMO degree per user antenna for uplink and downlink. Examples of uplink-specific parameters may include uplink channel conditions estimated by the eNB, noted as ULCQI(1 . . . Nue). Examples of uplink buffer status reports from UEs, noted as UBacklog [1 . . . Nue] may include the amount of data backlogged for uplink in each UE of such an eNB.
A network's criteria expressed in these parameters is combinatorially explosive and thus may not be possible to optimize within each frame, subframe or slot. Additionally, the schedule and its consequences may not reflect the value that a user would place on the opportunity to communicate. The network's criteria may reflect the complexity of managing the radio resources, such as traffic capacities (e.g. data rate), time delay, timing jitter, connection reliability, CQI, and other parameters known in the art as quality of service (QoS). When network capacities are exceeded, criteria for allocating radio resources may not reflect the tradeoffs a user might make regarding specific radio resources. Thus although the eNB includes parameters for Backlog and QoS in its allocation decisions, such an eNB may not have a parameter expressing user's current quality of experience (QoE). For example, such conventional scheduling does not consider whether a user is browsing a map for a future trip, or if the user is using the map to decide whether to exit the Interstate highway in 1 mile when traveling at 70 mph. The eNB has the basic service requests (BSR) of each eNB, but not the use context. There may be little or no dialog between a network and a given mobile node or collection of such nodes to express such context-sensitive degree of need for a given radio resource.
Methods for network-based radio resource allocation include the water filling algorithm and the Frank Kelly algorithm. The water filling algorithm applies available resources to the user (e.g. mobile node) most in need if resources are available when the node requests them. The Frank Kelly algorithm finds a data rate per user that is optimal with respect to a decision parameter which is the product of the utilities of the individual users. For the mathematics of the Frank Kelly algorithm to yield an optimum, users' needs must be elastic. That is, the utility function of user j on node j must be continuous, e.g. such as expressed in a number between 0 and 1.0. The value that a user may place on communication via a smart phone may not be elastic. In addition, the Frank Kelly algorithm may place significant traffic into the network for its iterative value estimate to converge to a set of utilities per user, Uj for 1≦j≦N, for N users, each of which may employ a mobile node.
An improved Frank Kelly algorithm may distribute decision making and may improve utility metrics. However, the water filling and Frank Kelly algorithms do not reflect the differences in value of network access among different applications. Such different applications may have different requirements for guarantee of bit rate, different likelihood of maintaining and/or losing connection, and with smaller or larger physical extent (e.g. the difference between a Wi-Fi network, which may have an extent of 30 meters, and a cellular network, which may have an extent of 2 kilometers). Improved Frank Kelly and water filling algorithms do not reflect differences in the value that a user may ascribe to a specific application or class of application at a point in space and time that may be different from that in another point in space and time.
These systems, methods, objects, features, and advantages of the present invention will be apparent to those skilled in the art from the following detailed description of the preferred embodiment and the drawings. All documents mentioned herein are hereby incorporated in their entirety by reference.